Registration Form

Grids '98: Designing, Building, and Using a National-Scale Grid

Grids '98 is a workshop focused on the technical and policy issues associated with the construction and use of a national-scale high-performance "grid."

Numerous Federal programs are by now underway or being planned to develop, prototype, integrate, and/or use various aspects of national-scale high-performance grid capabilities: the NSF PACIs, ASCI DISCOM, DOE2000, DOE's Strategic Simulation Initiative, NASA's Information Power Grid; also NGI, DARPA's Quorum, and Internet 2. These efforts promise collectively to enable revolutionary new approaches to science and engineering, based for example on tele-immersion and collaborative engineering; distributed supercomputing; tele-operation of scientific instruments; and knowledge synthesis from very large distributed databases.

However, significant technical and political obstacles must be overcome before a national infrastructure is in place and in routine use. The Grids '98 workshop is intended as a forum for discussion of important common concerns: purpose, architecture, practical deployment issues, long-term research problems, infrastructure requirements, opportunities for cooperation, etc. The goal is to identify concrete action items that can accelerate progress towards an integrated national infrastructure, and to take steps towards establishing a national grid community.

A new book collects papers presented and reviewed at an earlier workshop, held in September 1997; we will have copies of this available at Grids '98:

The Grid: Blueprint for a New Computing Infrastructure

	I. Foster and C. Kesselman (Eds)
	Morgan Kaufmann Publishers, 1998
	http://www.mkp.com/grids

Participants will include representatives of all the various national grid efforts. The primary product will be a white paper summarizing what is learned in the workshop and implications for future actions.

The meeting will be held at the Drake Hotel in Chicago on July 27 and 28, 1998 (see below for accomodation details). These are the two days before the IEEE High-Performance Distributed Computing conference, also held at the Drake (July 29-31: http://www.mcs.anl.gov/hpdc7), which many of you may also find of interest.

Sunday July 26th

Monday July 27th

Tuesday July 28th

Session/Breakout Group Charges

1. 1. Why Grids? Future Grid Applications

   The goal of this panel/breakout group is to characterize future grid applications (both short term and long term) in terms of both their functionality and their implications for infrastructure, tools, etc. The following specific questions are suggested:

   • In what areas can we expect major benefits?
   • What are major application areas that have not yet been exploited?
   • Can we quantify benefits?
   • Can we characterize required grid functionality, in terms of scale, scope, performance, etc.?

2. 2. Building the Grid: Technical and Policy Issues

   The goal of this panel/breakout group is to discuss the short-term steps that need to be taken to build usable national-scale (and international-scale) grids: in particular, the policies, standards, and technologies required to allow the sharing of resources located at different institutions and paid for in different ways. The following specific questions are suggested:

   • What are the critical obstacles to large scale resource sharing?
   • What inter-institutional structures can be used to overcome these obstacles?
   • What approaches and mechanisms have worked in the past, or failed in the past?
   • What are the principal areas in which standards are important?
   • What is the potential role of policy negotiation mechanisms?

   Research Challenges and Directions

   The goal of this panel/breakout is to discuss the central problems that require future research: in particular, potential showstoppers that might require urgent action. The following specific questions are suggested:

   • What are the principal obstacles to progress in the short term, medium term, and long term? (I.e., where is research required, now and in the future?)
   • What level of research activity do we currently see in each area? (I.e., where are there gaps in research portfolio?)
   • What research activities can be proposed?

3. Grid Experiences: What Have We Learned?

   The goal of this panel is to summarize what we've already learned from gigabit testbeds, I-WAY, and other attempts to build grids at scale. The following specific questions are suggested:

   • What were the biggest successes, failures, obstacles, etc.?
   • What problems can be regarded as "solved"?
   • What problems remain open?
   • What lessons should we draw from past experiences for future grid efforts?

4. What Can't the Grid Do?

   The goal of this panel is to identify applications for which the Grid is *never* going to be suitable. The intention is to force people to think about the operating parameters of grids: to examine the edges, not just the sweet spots. Hence, the following specific questions are suggested:

   • What classes of applications cannot be expected to benefit from grid technologies, perhaps because all computations are tightly coupled (can we define "tightly coupled"?)?
   • What classes of grid-oriented applications can be addressed quite adequately with today's (or next year's) Internet technologies?
   • What classes of applications seem promising but appear to require a technology that we do not expect to see any time soon?